I''m running a fairly recent version of xen: [kmacy@xen0 xen-3.1-testing.hg]$ hg tip changeset: 15547:f4dc0d3dfb0f tag: tip user: Keir Fraser <keir.fraser@citrix.com> date: Sat Dec 08 17:23:05 2007 +0000 summary: Update version tag to 3.1.3-rc1-devel (nearly ready for -rc1). I''m currently doing bring-up inside of a VMware VM. The VM resets itself with a stack fault when I execute a MMUEXT_PIN_L1_TABLE. Has any bug that could be causing this been fixed recently? If not could someone do me a quick favor and try out: http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz All it takes for me is ''xm create xmexample1'' to cause dom0 to be reset. Thanks. -Kip _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Actually, it turns out that it occurs slightly further up. Trying to add a page directory with only the valid flag set seems to make Xen unhappy (i.e. system reset). Setting the R/W flag doesn''t seem to work as it complains: (XEN) mm.c:1477:d1 Bad L3 flags 2 In any event I''ll figure out what the right incantation is, but it appears you may have a legitimate hole at the moment. -Kip On Dec 24, 2007 4:39 PM, Kip Macy <kip.macy@gmail.com> wrote:> I''m running a fairly recent version of xen: > > [kmacy@xen0 xen-3.1-testing.hg]$ hg tip > changeset: 15547:f4dc0d3dfb0f > tag: tip > user: Keir Fraser <keir.fraser@citrix.com> > date: Sat Dec 08 17:23:05 2007 +0000 > summary: Update version tag to 3.1.3-rc1-devel (nearly ready for -rc1). > > I''m currently doing bring-up inside of a VMware VM. The VM resets > itself with a stack fault when I execute a MMUEXT_PIN_L1_TABLE. > > Has any bug that could be causing this been fixed recently? If not > could someone do me a quick favor and try out: > > http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz > > All it takes for me is ''xm create xmexample1'' to cause dom0 to be reset. > > > Thanks. > > -Kip >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
If this is a PAE system then L3 entries only have the present/valid bit -- there''s certainly no r/w bit. Try upgrading to tip of xen-3.1-testing (c/s 15564) as that has definitely passed regression tests (which involves creating quite a number of domains). Perhaps we have a bad interaction with VMware. -- Keir On 25/12/07 01:40, "Kip Macy" <kip.macy@gmail.com> wrote:> Actually, it turns out that it occurs slightly further up. Trying to > add a page directory with only the valid flag set seems to make Xen > unhappy (i.e. system reset). Setting the R/W flag doesn''t seem to work > as it complains: > > (XEN) mm.c:1477:d1 Bad L3 flags 2 > > In any event I''ll figure out what the right incantation is, but it > appears you may have a legitimate hole at the moment. > > -Kip > > > On Dec 24, 2007 4:39 PM, Kip Macy <kip.macy@gmail.com> wrote: >> I''m running a fairly recent version of xen: >> >> [kmacy@xen0 xen-3.1-testing.hg]$ hg tip >> changeset: 15547:f4dc0d3dfb0f >> tag: tip >> user: Keir Fraser <keir.fraser@citrix.com> >> date: Sat Dec 08 17:23:05 2007 +0000 >> summary: Update version tag to 3.1.3-rc1-devel (nearly ready for -rc1). >> >> I''m currently doing bring-up inside of a VMware VM. The VM resets >> itself with a stack fault when I execute a MMUEXT_PIN_L1_TABLE. >> >> Has any bug that could be causing this been fixed recently? If not >> could someone do me a quick favor and try out: >> >> http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz >> >> All it takes for me is ''xm create xmexample1'' to cause dom0 to be reset. >> >> >> Thanks. >> >> -Kip >> > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 25/12/07 00:39, "Kip Macy" <kip.macy@gmail.com> wrote:> Has any bug that could be causing this been fixed recently? If not > could someone do me a quick favor and try out: > > http://www.fsmware.com/xenofreebsd/stackfault/xen.tgzI''ve downloaded it. I assume I try to create this with PAE Xen and PAE dom0? I''ll try a bit later this week. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Correct. Just start it up on a PAE xen. As it turns out *any* in-place updates to the L3 - even putting a zero where there is currently a zero - will cause a system reset. At least on VMWare Fusion. I''ve worked around it for the moment by allocatng a new L3 and then switching to that. -Kip On 12/25/07, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote:> On 25/12/07 00:39, "Kip Macy" <kip.macy@gmail.com> wrote: > > > Has any bug that could be causing this been fixed recently? If not > > could someone do me a quick favor and try out: > > > > http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz > > I''ve downloaded it. I assume I try to create this with PAE Xen and PAE dom0? > I''ll try a bit later this week. > > -- Keir > > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
When does the crash happen for you? While the domU boots? Or during domain building? -- Keir On 25/12/07 10:16, "Kip Macy" <kip.macy@gmail.com> wrote:> Correct. Just start it up on a PAE xen. > > As it turns out *any* in-place updates to the L3 - even putting a zero > where there is currently a zero - will cause a system reset. At least > on VMWare Fusion. I''ve worked around it for the moment by allocatng a > new L3 and then switching to that. > > -Kip > > > > On 12/25/07, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote: >> On 25/12/07 00:39, "Kip Macy" <kip.macy@gmail.com> wrote: >> >>> Has any bug that could be causing this been fixed recently? If not >>> could someone do me a quick favor and try out: >>> >>> http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz >> >> I''ve downloaded it. I assume I try to create this with PAE Xen and PAE dom0? >> I''ll try a bit later this week. >> >> -- Keir >> >> >>_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Early in boot. -Kip On 12/25/07, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote:> When does the crash happen for you? While the domU boots? Or during domain > building? > > -- Keir > > On 25/12/07 10:16, "Kip Macy" <kip.macy@gmail.com> wrote: > > > Correct. Just start it up on a PAE xen. > > > > As it turns out *any* in-place updates to the L3 - even putting a zero > > where there is currently a zero - will cause a system reset. At least > > on VMWare Fusion. I''ve worked around it for the moment by allocatng a > > new L3 and then switching to that. > > > > -Kip > > > > > > > > On 12/25/07, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote: > >> On 25/12/07 00:39, "Kip Macy" <kip.macy@gmail.com> wrote: > >> > >>> Has any bug that could be causing this been fixed recently? If not > >>> could someone do me a quick favor and try out: > >>> > >>> http://www.fsmware.com/xenofreebsd/stackfault/xen.tgz > >> > >> I''ve downloaded it. I assume I try to create this with PAE Xen and PAE > dom0? > >> I''ll try a bit later this week. > >> > >> -- Keir > >> > >> > >> > > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 25/12/07 10:16, "Kip Macy" <kip.macy@gmail.com> wrote:> Correct. Just start it up on a PAE xen. > > As it turns out *any* in-place updates to the L3 - even putting a zero > where there is currently a zero - will cause a system reset. At least > on VMWare Fusion. I''ve worked around it for the moment by allocatng a > new L3 and then switching to that.I can''t reproduce this with xen-3.1-testing.hg c/s 15547 on real hardware. Although the domain does crash in a fairly intricate and interesting way. I''ve appended the Xen logging of the domain crash. I assume your ''whole machine'' crash is simply inaccurate emulation of native behaviour by VMware Fusion. -- Keir #8, proc0kstack=3227074560 xen_setbootenv(): cmd_line=''root=/dev/hda1 ro 4'' gdt=0xc0596000 PTmap=0xbf800000 (XEN) Unhandled page fault in domain 1 on VCPU 0 (ec=0000) (XEN) Pagetable walk from bfe02cb4: (XEN) L3[0x002] = 00000000aec46001 0000059c (XEN) L2[0x1ff] = 0000000000000000 ffffffff (XEN) domain_crash_sync called from entry.S (ff17f5a2) (XEN) Domain 1 (vcpu#0) crashed on cpu#3: (XEN) ----[ Xen-3.1.3-rc1-devel x86_32p debug=y Not tainted ]---- (XEN) CPU: 3 (XEN) EIP: e019:[<c040a7e7>] (XEN) EFLAGS: 00000206 CONTEXT: guest (XEN) eax: 000c0596 ebx: 00000098 ecx: deadbeef edx: deadbeef (XEN) esi: c04f2184 edi: 00000000 ebp: c0595d84 esp: c0595bf4 (XEN) cr0: 8005003b cr4: 000026f0 cr3: 00260d20 cr2: bfe02cb4 (XEN) ds: e021 es: e021 fs: e021 gs: e021 ss: e021 cs: e019 (XEN) Guest stack trace from esp=c0595bf4: (XEN) 00000000 c040a7e7 0001e019 00010006 c0492b92 bf800000 00000000 00000001 (XEN) 00000000 00000000 00000001 00000000 00000000 00000000 00000000 00000000 (XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 (XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On Dec 26, 2007 4:37 AM, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote:> On 25/12/07 10:16, "Kip Macy" <kip.macy@gmail.com> wrote: > > > Correct. Just start it up on a PAE xen. > > > > As it turns out *any* in-place updates to the L3 - even putting a zero > > where there is currently a zero - will cause a system reset. At least > > on VMWare Fusion. I''ve worked around it for the moment by allocatng a > > new L3 and then switching to that. > > I can''t reproduce this with xen-3.1-testing.hg c/s 15547 on real hardware. > Although the domain does crash in a fairly intricate and interesting way. > I''ve appended the Xen logging of the domain crash. I assume your ''whole > machine'' crash is simply inaccurate emulation of native behaviour by VMware > Fusion.Excellent, I''m glad to hear that it is a VMWare Fusion bug (as opposed to an easily triggered Xen bug). PAE isn''t expected to work just yet, so *a* crash is expected. Thanks. -Kip> > -- Keir > > #8, proc0kstack=3227074560 > xen_setbootenv(): cmd_line=''root=/dev/hda1 ro 4'' > gdt=0xc0596000 > PTmap=0xbf800000 > (XEN) Unhandled page fault in domain 1 on VCPU 0 (ec=0000) > (XEN) Pagetable walk from bfe02cb4: > (XEN) L3[0x002] = 00000000aec46001 0000059c > (XEN) L2[0x1ff] = 0000000000000000 ffffffff > (XEN) domain_crash_sync called from entry.S (ff17f5a2) > (XEN) Domain 1 (vcpu#0) crashed on cpu#3: > (XEN) ----[ Xen-3.1.3-rc1-devel x86_32p debug=y Not tainted ]---- > (XEN) CPU: 3 > (XEN) EIP: e019:[<c040a7e7>] > (XEN) EFLAGS: 00000206 CONTEXT: guest > (XEN) eax: 000c0596 ebx: 00000098 ecx: deadbeef edx: deadbeef > (XEN) esi: c04f2184 edi: 00000000 ebp: c0595d84 esp: c0595bf4 > (XEN) cr0: 8005003b cr4: 000026f0 cr3: 00260d20 cr2: bfe02cb4 > (XEN) ds: e021 es: e021 fs: e021 gs: e021 ss: e021 cs: e019 > (XEN) Guest stack trace from esp=c0595bf4: > (XEN) 00000000 c040a7e7 0001e019 00010006 c0492b92 bf800000 00000000 > 00000001 > (XEN) 00000000 00000000 00000001 00000000 00000000 00000000 00000000 > 00000000 > (XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 > 00000000 > (XEN) 00000000 00000000 00000000 00000000 00000000 00000000 00000000 > 00000000 > > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel